Upload
others
View
6
Download
0
Embed Size (px)
Citation preview
Parc Mediterrani de la Tecnologia
Edifici ESAB
Avinguda del Canal Olímpic 15
08860 Castelldefels
MEASURING AND MANAGING CONSUMERS'
RISK PERCEPTION TOWARDS FOOD RELATED
ISSUES
Wisdom Dogbe and José M. Gil CREDA-UPC-IRTA
Castelldefels (Barcelona)
To start with New food technologies available
Relation between diet and health
Food scares
Increasing consumers concerns about food safety and health issues
Risk perception
Changing attitudes
Changing behavior
Policy makers
Regulation
Industry
Brands, HACCP
Health claims
Social scientists
Research
multidisciplinary
Mass media
Food
Food borne diseases Diseases related to diet
To start with
Food
Food borne diseases Diseases related to diet
To start with
To start with
5 Angulo and Gil (2007). Food Quality and Preference
Consumers’ perceptions of risk and benefit associated with GM food
Individual attributes and
values (general attitudes)
Information
Consumers’ attitude towards GM food (acceptance
or rejection)
Consumers’ intentions towards
GM food
Consumer purchase behaviour of GM food
Price and
other factors
Knowledge of product and
process
(As special attributes)
Trust
Costa, Gil, Traill (2008). Food Policy
To start with
Outline of the presentation
• Risk Perception & Risk Aversion
• Linking Risk Perception & Risk Attitudes
• Measuring Risk Attitudes: Simple Methods
• Expected Utility Framework
• Prospect Theory Framework
• Measuring Risk Attitudes: Complex Method
• Empirical Application
• Estimating Prospect Theory Parameters
• Preliminary Results 1
• Relating Risk Attitudes & BMI
• Preliminary results 2
7
Risk Perception & Risk Aversion 1
• In Business:
• A person’s risk propensity influences evaluation of
risky situation.
• Risk propensity may impact risk perception
(Brockhaus 1980; Vlek and Stallen 1980).
• Risk propensity has an inverse effect on risk
perception ( Keil et al.,2000; Forlani et al. (2002).
8
Risk Perception & Risk Aversion 2
• Food Safety
• Less risk averse consumers perceive food safety
risk to be very low in case of an outbreak
(Schroeder et al. 2017; Weller, Andrea and Caleb
(2012).
• Consumption only reduces when the risk
perception is relatively high
• Consequently, less risk averse people rarely
reduce consumption
9
Linking Perception & Attitudes 1
Based on results from Schroeder et al. (2017) Weller, Andrea and Caleb (2012)
10
20% Health Risk
Risk Lover
Perceive lower danger
Risk Averse
Perceive higher danger
Linking Perception & Attitudes 2
So,
• Risk attitudes negatively affect risk perceptions
• Risk attitudes are inherent to consumers
• Risk perceptions are more conjectural (measurement is ad hoc and case
specific) and depend on information, the technology itself, mass media or
social networks and risk attitudes
• Other presentations on risk perception
• We focus on risk attitudes and, more specifically, how to
measure them?
11
Measuring Risk Attitudes 1 Respondents give a global assessment of their willingness to take
risks.
Framing
“How do you see yourself: are you generally a person who is fully prepared
to take risks or do you try to avoid taking risks”
Respondents are assessed on the scale of 0 - 10:
0 => not at all willing to take risks
10 => very willing to take risks
(Dohmen et al., 2011)
12
Measuring Risk Attitudes 2
13
Applied in
Neuroscience (Fecteau et al., 2007)
Drug addiction (Bornovalova et al.,
2005) and
Psychopathology (Hunt et al., 2005).
WEAKNESS
It is not clear if risk preferences extend
to other domains
Requires a computer and multiple
trials to implement
1. Balloon Analogue Risk Task (Lejuez et al., 2002)
Experimental Methods - Simple Computer Screen
Measuring Risk Attitudes 3
14
Healthy &
Safety 8
items
Ethical 8
items
Recreational 8
items
Social 8
items
Gambling 4
items
Investment 4
items
very unlikely unlikely not sure likely very likely
1 2 3 4 5
Simple to understand method
Critics:
Questionnaires are not incentivized:
Hence, elicited risk preferences
may partially reflect an
individual’s true attitudes toward
risk
Hanoch et al. (2006) used
the DOSPERT to
demonstrate the domain-
specific nature of risk
preferences.
Preference (X) = a*Expected Benefit (X) + b*Perceived Risk (X) + c
Measuring Risk Attitudes 4
15
Endowment ($5)
Safe account
($5-X)
Risky investment
(50%:2.5*X; 50%:0)
The Investment game risk-elicitation method: from certain to uncertain
3. The Gneezy and Potters method
Critics:
Does not distinguish between risk-
seeking and risk-neutral
preferences
Used to elicit myopic loss aversion in the financial
decisions among
• students (Gneezy and Potters, 1997),
• professional traders (Haigh and List, 2005)
Compare gender differences in risk attitudes
(Charness and Gneezy, 2012).
Risk preferences of bridge players How much would you like
to invest (X)
Measuring Risk Attitudes 5
16
4. Eckel-Grossman Task
Results correlated significantly with those elicited through the other methods (Reynaud and
Couture, 2012)
Produced significantly less noisy estimates of risk preferences more than complex ( Dave et al.,
2010)
Relatively easy for individuals to understand
Critics:
it cannot differentiate between different degrees of risk-seeking behaviour
Expected Utility
17
• Preferences towards risky choices are represented by utility function (ordinal, not cardinal) U(a)
– von Neumann Morgenstern utility function
• Decisions are made to maximize expected utility EU(a)
– E is the expectation operator based on subjective probability distributions of a
• Independence assumption violated (assumption of linearity in probabilities may not hold).
• Risk preference characterized by expected utility (EU) assume that,
– Risk aversion is the sole parameter for determining the curvature of the utility function.
Prospect Theory 1
• In Prospect Theory (PT) losses are valued more heavily than gains
–Presence of loss aversion
• PT postulate
– risk aversion for gains, concave utility function
– risk seeking to avoid losses, convex utility function
18
Prospect Theory 2
19 Loss aversion and Risk Aversion in Prospect Theory
Prospect Theory 3
• In PT the shape of the utility function is jointly determined by
– risk aversion,
– loss aversion (which measures one’s sensitivity to loss compared to gain),
– and nonlinear probability weighting (the individual tendency of overweighting small (large) probabilities and underweighting large (small) probabilities).
20
Prospect Theory 4
21 Probability Weighting in the Prospect Theory
Prospect Theory 5
22
• The MPL was designed to allows the researcher to
estimate models that
– nest both EU and PT
• Also MPL allows the results from the experiment to
determine whether EU or PT better fits the data.
Measuring Risk Attitudes 1
23
Complex Method: Holt–Laury measure of risk aversion
Participants are typically informed that one
decision will be selected at random and the
chosen gamble will be played for real.
Subjects are then paid according to that
outcome.
Study relationship between
• risk aversion and cognitive ability
(Dohmen et al. 2010)
Measuring Risk Attitudes 2
24
• Modified/Double Multiple Price List Method all 3
prospect theory parameters
– concavity,
– loss aversion,
– and weighting function parameters.
Measuring Risk Attitudes 3
25
• Modified MPLs
Critics:
Most subjects will fail to understand the
procedure
- reduces the reliability of estimates
Some participants may make inconsistent
decisions
- Solved by imposing strict monotonicity and
enforcing transitivity.
No consensus about the application in other
domain
Applied to examine the preferences of
Vietnamese villagers (Tanaka et al. 2010)
Empirical Application 1
26
• Many researchers have applied the MPL to elicit risk preferences (Dohmen et al. 2011; Charness and Viceisza, 2011; Anderson and Mellor, 2009; Lonnqvist et al., 2011; Reynaud and Couture, 2012; Dave et al., 2010).
• Applied to sample population that include:
– Students, Farmers, rural villagers and residents
• No study yet on consumer behaviour
– area of food/health policy
Empirical Application 2
27
• We study risk attitudes of consumers by
– Analysing correlation between risk aversion and BMI
• We used the
– cumulative prospect theory (Kahneman and Tversky, 1979; Tversky and Kahneman, 1992)
– and the one-parameter form of Prelec’s (1998) axiomatically derived weighting function
Estimating Prospect Theory Parameters
28
• Under the PT, Utility function is modelled by
– 𝑃𝑇 𝑥, 𝑦; 𝑝 =
𝑝𝑣 𝑥 + 1 − 𝑝 𝑣(𝑦); x > y > 0 or x<y<0
w(𝑝)𝑣 𝑥 + 𝑤 𝑝 𝑣(𝑦); x<0<y
• Value Function:
𝑣 𝑥 = 𝑥σ 𝑓𝑜𝑟 𝑥 ≥ 0
−𝜆 −𝑥σ 𝑓𝑜𝑟 𝑥 < 0
• Weighting function: 𝑤 𝑝 = 𝑒𝑥𝑝 −(− ln 𝑝)𝛾
Estimating Prospect Theory Parameters
29
• Series 1 and series 2 were used to estimate
– the curvature of the utility function (σ)
– and the nonlinear probability weighting parameter (γ)
for each respondent
• Using σ, γ estimated from above and the switching
point of series 3,
• we estimated the loss aversion parameter (𝜆)
Preliminary Results 1
30
• Average risk aversion parameter to be 0.5875,
– Consumers are in general risk averse.
• The average loss aversion parameter is 3.67,
– In general consumers are loss averse.
• Average of the probability weighting parameter is 0.69,
– In general consumers have the tendency to overweight low probabilities.
• Since σ is not equal to 1 and γ is not equal to 1
– We reject expected utility framework
Relating Risk Attitudes and BMI
31
• Past studies suggest that
– increase in risk aversion will lead to a decrease in BMI,
– an increase in loss aversion will lead to an increase in
individual´s BMI.
• As such we postulate that risk aversion and loss
aversion correlate with an individual´s BMI.
Relating Risk Attitudes and BMI
32
• We estimate linear regression model (with robust standard errors):
– relate risk preference parameters to BMI and other socioeconomic characteristics
• σ𝑖 = 𝛿0 + 𝛿1𝐵𝑀𝐼𝑖 + 𝛿3γ𝑖 + 𝛿4𝑔𝑒𝑛𝑑𝑒𝑟𝑖 + 𝛿5𝐴𝑔𝑒𝑖 + 𝛿6𝑚𝑎𝑟𝑖 + 𝛿7 + 𝑝𝑟𝑖𝑚𝑖 +
𝛿8𝑠𝑒𝑐𝑖
• Mar implies the person is married
• prim is 1 if the individual´s highest level of education is primary,
• sec is 1 if the individual´s highest level of education is secondary education and 0 if otherwise.
Preliminary Results 2
33
Risk Aversion Loss Aversion
BMI 0.01* 0.03
Age -0.01** 0.06***
Probability weighting 0.05 -2.12
Married -0.05 0.79
Gender 0.08 -0.73
Primary education 0.09 -0.10
Secondary education 0.10** -0.85
Constant 0.50* 2.05
*,**,*** respresent significant at 10%, 5%, 1%, respectively.
Obese persons are
less risk averse
Older People
are less loss
averse
Secondary school leavers are
less risk averse than university
graduates
Older People are
more risk averse
Preliminary Results 2
34
Risk Aversion Loss Aversion
BMI 0.01* -0.125**
Age -0.010*** 0.042
Probability weighting 0.049 -0.733
Married -0.045 -0.088
gender 0.081 -0.653
Primary education 0.094 -0.673
Secondary education 0.096** 0.214
Constant 0.499* 4.327**
• We performed a robustness check by
• excluding all individuals who did not switch from A to B or chose option B
throughout.
*,**,*** respresent significant at 10%, 5%, 1%, respectively.
Obese persons are
less risk averse
Older People are more
risk averse
Secondary school leavers
are less risk averse than
university graduates
Increase in BMI
increases loss
aversion
Measuring Risk Attitudes
35
Thank you